Prediction of the physical properties of degradable plastics using ANFIS / Syamsiah Abu Bakar
Degradable plastic is produced by combining different percentages of additives such as oil palm biomass (OPB), palm oil (PO) and starch with polyethylene (PE). Currently, experiments are carried out in laboratories to determine the formulation of degradable plastics with the most bioactive component...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 语言: | English |
| 出版: |
2012
|
| 在线阅读: | https://ir.uitm.edu.my/id/eprint/79013/1/79013.pdf |
| 标签: |
添加标签
没有标签, 成为第一个标记此记录!
|
| 总结: | Degradable plastic is produced by combining different percentages of additives such as oil palm biomass (OPB), palm oil (PO) and starch with polyethylene (PE). Currently, experiments are carried out in laboratories to determine the formulation of degradable plastics with the most bioactive components and desirable physical properties measured by melt How index (MFI), melting point (MP) and density. The procedure is time consuming and costly. Therefore, a different approach is required to minimize the time consumption, production cost and labour cost. In this research, an Adaptive Neuro-Fuzzy Inference System (ANFIS) and linear regression (LR) models have been developed and utilized to predict the physical properties of degradable plastics. The prediction accuracy of ANFIS and LR models are assessed by comparing simulated results with the actual lab results using root mean square error (RMSE), correlation coefficient (/?), coefficient of determination (R2 ) and adjusted R square (R2 ). ANFIS and LR models are found to have compatible prediction performances. The findings show that ANFIS model is capable of determining the desirable input-output relationships in degradable plastic production as reflected by the small RMSE values, high R and R2 values. Furthermore the difference between R2 and R2 values is small which indicates that the addition of new variable does not contribute to the over fitting of ANFIS model. It was also found that different membership functions used in ANFIS has an impact on ANFIS prediction performance. The developed ANFIS model can be used as a guide for the production of degradable plastic with some intended physical requirements. |
|---|